DEDUPED Hot AI repos. Real use cases. What people build. Get the Monday 5

How to load PDFs, Markdown and CSV files into a graph you can question with Helix Foundry

Upload documents and table files into Foundry so they join the same graph as your databases.

Short answer: Run Foundry with docker compose up -d --wait; then upload Markdown, text or PDF documents and CSV or TSV table files; let the agent draft types and links that tie the documents to your other data. All steps.

Who it is forA team whose key facts live in PDFs, notes and spreadsheets rather than a database.

The problemFacts in documents never connect to the records they talk about.

The outcomeDocuments and table files loaded as part of one graph, with answers quoting the passage they came from.

What you need first

  • Docker with Compose v2 to run locally
  • Optional: ANTHROPIC_API_KEY or OPENAI_API_KEY in the shell for Claude or OpenAI
  • Node 24 and pnpm only for developing the site

Steps

  1. Run Foundry with docker compose up -d --wait
  2. Upload Markdown, text or PDF documents and CSV or TSV table files
  3. Let the agent draft types and links that tie the documents to your other data
  4. Load and explore as a graph or as records
  5. Ask a question and check the quote each answer cites

Prompt for your agent

Use HelixDB/helix-foundry to load <folder of PDFs and CSVs>, draft an ontology that links them, and answer <question> with quotes from the source documents.

Caveat. This card is written from the HelixDB/helix-foundry README, not from a test run by Deduped. Before you start, make sure you have Docker with Compose v2 to run locally. Check the repo for newer instructions before you rely on these steps.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for HelixDB/helix-foundry and "Load PDFs, Markdown and CSV files into a graph you can question"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer